In this paper, we combine a ventricular kinematic model and an ultrasound simulation model in order to simulate the
echocardiographic imaging process. In addition to its capability to generate raw RF data, when compared to previous
echocardiography simulation models, the result achieves more realistic B-Mode images. Several echocardiography
parameters were taken into account including central frequency, apodization, number of elements in the array, speed of
sound, and number of scatterers. The proposed improvements are due to the use of a shift-variant Point Spread Function
(PSF) and more accurate cardiac motion assumptions. One attribute of the simulator is also that it provides the groundtruth
vector field of actual "ventricular deformations'' which may be used to strictly validate motion estimation and
myocardial elastography algorithms.
The paper presents the first application of optical flow to normalized data in piece-wise segments of RF images.
Different optical flow motion estimation techniques such as Lucas-Kanade, Horn-Schunck, Brox et al., Black and
Anandan, and Block Matching (BM) were applied to the simulated B-mode images and RF data. The estimated motion
fields from the RF data as well as the B-mode images were validated with the ground-truth motion fields derived from
The validation results show that the Brox et al. method performs better than other motion estimation techniques when
applied to B-Mode and RF data. Also, as intuitively expected, use of RF data results in more accurate displacement
fields than when B-mode images alone are used.